GRD Journals- Global Research and Development Journal for Engineering | Volume 1 | Issue 12 | November 2016

ISSN: 2455-5703

Region Growth Based Segmentation to Improve

the Porosity of Cu - (520%) W Composite
Preforms
P. Radha
Department of Computer Applications
Mepco Schlenk Engineering College, Virudhunagar 626 005, Tamil Nadu, India

Abstract
While preparing the composite preform in the powder metallurgy Lab, the various defects due to porosity, open crack and
residual stresses are possible. This may lead to poor life and strength of materials. It is difficult to predict the defects in the form
of pores physically in the powder metallurgy Lab. To simplify this kind of problem, the Scanning Electron Microscope (SEM)
images are generated from the powder composites and are segmented using region growth approach to find the distribution of
pores. Normally, the composite preforms are being produced through various processes like mechanical milling, mixing,
compaction, sintering and hot extrusion. In this study, Cu(520%) W composite preforms, with a preform density of 94% are
prepared. The pore size in term of coverage area, perimeter during different sintering atmospheres are derived. Further, the
porosity is reduced during extrusion process. The results of SEM images are compared before and after sintering and extrusion
process. This kind of work will aid the manufacturing process of material parts in predicting their strength and life time.
Keywords- Image segmentation, Region growing, composites, porosity analysis

I. INTRODUCTION
The digital image processing techniques are being applied to automate all the processes of different applications [1]. The digital
images are manipulated to get more information about the properties of materials. While preparing powder composites in the
Powder Metallurgy (P/M) Lab, the various detects due to porosity, open crack and residual stresses are possible, which may lead
to poor life time and strength of materials. The pores are possible in various stages. This will affect the strength and lifetime of
material products. It is very difficult to measure the pores physically in the Lab. To overcome these problems, the SEM images
of these powder composites are generated and they are processed to calculate the distribution of pores as well as the contribution
of each powder material in the powder composite. In this study the Coppertungsten (CuW) composites are considered to
support the manufacturing of material products since the CuW composites are extensively used for their superior strength at
high temperature and having wear resistance for electrical discharge, electrode materials, relay blades and electrical contact
supports [2].
In this study the image processing techniques are applied to study the defects of materials in the form of SEM images,
since SEM is a powerful technique in the examination of materials. And also the digital image processing technique like
segmentation can classify the defect areas of the materials easily. The SEM images are used widely in metallurgy, geology,
biology and medicine. Also these SEM images are used to find the pores and the contribution of various materials. The material
products are being manufactured through various processes like mechanical milling, mixing, compaction, sintering and hot
extrusion. Cu(520%) W composite preforms, with a density of 94% were prepared in the lab. The defects occurred in various
stages are collected in the form of images. The pore size during different sintering atmospheres and the pore size reduction
during extrusion were studied.

II. LITERATURE SURVEY

Shwetabh Singh (2013) showed that the edges obtained in the images can be used for classification of particles, determining
sizes & shapes & also distinguishing particles in agglomerates more precisely[3]. Azmi Tawfik Alrawi et al. (2012) studied that
the proportion of porosity on the surface of CdS thin film less in higher annealing temperature, because the annealing works on
recrystallized grains of the thin film so the white areas are growing and which represents the surface of thin film and the other
hand, the black areas less than which represent pores, cracks and defects on the surface of the thin membrane [4]. Meena et al.
(2014) analysed SEM images with pore characteristics [5]. Azmi Tawfik Alrawi et al. (2012) obtained the proportion of porosity
on the surface of CdS thin film [4] Gary Chinga ( 2002) used image analysis techniques to handle micro structures of paper [6].
Manuel F.M. Costa (2004) applied image processing techniques to the characterization of Nano structures [7]. M. G. CortinaJanuchs et al. (2011) detected pore space in CT soil images [8]. Aref Naimzad et al. (2014) studied comparatively on Mechanical
and Magnetic properties of porous and Nonporous Film-shaped Magnetorheological Nan composites [9]. Ahmet H. Aydilek et

All rights reserved by www.grdjournals.com

148

Region Growth Based Segmentation to Improve the Porosity of Cu - (520%) W Composite Preforms
(GRDJE/ Volume 1 / Issue 12 / 025)

al. (2002) measured fiber thickness and pore opening sizes in a cross-sectional image a woven geotextile. [10] Shawn Zhang,
Robert E. Klimentidis et al. (2011) used image processing techniques for porosity and permeability analysis of SEM 3D images
[11].
In the previous research, mostly the image processing techniques were applied to the different images other than SEM
images related to the study of material properties. This approach used the SEM images to identify and rectify the defects without
destructing the materials.

III. EXPERIMENTAL DETAILS

5-20% of tungsten powders were mixed with copper to prepare different categories of preforms like Cu-0W, Cu-5W,Cu-10W
and Cu20W . The mixing was done using by a suitable punch and die set assembly of UTM, of 1 MN capacity. The composites
with 20 mm diameter were prepared. Compacting pressure was applied gradually, and it was 1.2 GPa for all the specimens. The
Graphite was used as lubricate while preparing the composites. The initial density 87% was maintained by accurately controlling
the mass and observing the compacting pressure employed. After the compaction, the compacts were immediately taken out
from the die set assembly and loaded into the sintering furnace for sintering. The sintering process was performed using different
conditions, namely, normal atmospheric air and argon atmosphere at various sintering temperatures such as 750C, 800C,
850C, for a heating period of 1 h using an electric muffle furnace. The sintering temperature was increased to get good bonding
and the sintering temperature was randomly increased to 750C. Again this was increased from 750C to 850C and the
hardness was improved after sintering. Sintering of Cu at 850C under normal atmospheric conditions had the presence of large
sized visible pores. Then sintered composite of Cu at 800C using argon atmospheric conditions produced small size visible pores
. Finally the different categories of composites like Cu-0W, Cu-5W,Cu-10W and Cu20W were applied to the extrusion process
for improving the hardness further.
A. Methodology Adopted
Image is a collection of signals. These signals are processed to derive meaningful information. The image is defined as a 2D
functions (x, y), where x and y represents are the spatial coordinates and the amplitude off at any pair of coordinates (x,y) is
called the intensity or the gray level of the image at that point or pixel [1]. It is used in various domains. The images are
manipulated through various stages like Acquisition, prep-processing, image enhancement, image transformation, image
segmentation, image representation and Image recognition.
The first step in image processing system is Image Acquisition, where the optimal image (real world data) is converted
into array of numeric data. In the field of powder metallurgy, for collecting the images, SEM (Scanning Electron Microscope) is
used. It is a type of electron microscope that produces images of a sample by scanning it with a focused beam of electrons. The
electrons interact with atoms in the sample, producing various signals that can be detected and that contain information about the
sample's surface topography and composition. The electron beam is generally scanned in a raster scan pattern, and the beam's
position is combined with the detected signal to produce an image. SEM can achieve resolution better than 1 nm. Specimens can
be observed in high vacuum, in low vacuum, in dry conditions (in environmental SEM), and at a wide range of cryogenic or
elevated temperatures. The images are stored in gif format.
In image processing, segmentation is the core stage used to partition the image into various parts with respect to their
similarity. Segmentation is the process of extracting points, lines or regions, which are then used as inputs for complementary
tasks such as registration. Segmentation falls into two categories intensity based segmentation and region based segmentation.
The region based segmentation is a technique in which segmentation is carried out based on the similarities in the given region.
Region growing is a classification of region based segmentation. As the name implies, region growing is a procedure in which
pixels are grouped based on some predefined condition. It is a pixel-based image segmentation using initial seed points to
identify the regions of images. This method examines neighboring pixels of initial seed points and determines whether the pixel
neighbors should be added to the region. The process is iterated until the sufficient regions are identified. The initial region
begins as the exact location of these seeds. Region growing methods can provide the original images which have clear edges
with good segmentation results. The SEM images are generated for the Cu(520%) W composite preforms before and after
sintering process. The SIPNC_Cu_W Algorithm (SEM IMAGE POROSITY for nano composite of type: Coppertungsten) is
used to determine the porosity and its characteristics.
B.
1)

2)

Algorithm SIPNC_Cu_W
For each core process like sintering / extrusion of Manufacturing Department
Capture the SEM image at the starting of the process
Applied to image processing
Reading and Resizing the image
Removing noise of Image
Enhancing the images
Apply region based segmentation
Pick up an arbitrary point(x,y) of the image as seed

All rights reserved by www.grdjournals.com

149

Region Growth Based Segmentation to Improve the Porosity of Cu - (520%) W Composite Preforms
(GRDJE/ Volume 1 / Issue 12 / 025)

3)

Consider the neighbours of seed point using 4 or 8 connected method

Check each neighbouring pixel whether it is in the same region as (x,y)
Accept the neighbouring pixel if it satisfies the homogeneity property of a region.
Once the newly accepted is assigned into the region, its neighbours are checked using 4 or 8 connected method.
The segmentation is repeated until no more new pixels.
Finally the labels are assigned for each region
Calculate the number of regions of an Image, total area and total perimeter of the image. Also predict the texture properties
such as average intensity, average contrast and entropy information of an image.
4) If the pore size is larger, the sintering process is repeated with updated temperature.
5) For the resultant powder composite the image is generated and the above steps are repeated until the pore size gets reduced.
Further the pore sizes are recued further using extrusion process. The intensity of seed point and the threshold values are
reported in Table 3 for the SEM images acquired before / after the sintering and extrusion processes.

IV. RESULTS AND DISCUSSION

Figure 1 shows the composite preforms after sintering. Fig. 2(a) shows the SEM image of the sintering of Cu at 850C under
normal atmospheric conditions. It reveals the presence of large sized visible pores. However, Fig. 2(b) shows the SEM image of
sintered composite of Cu at 800C using argon atmospheric conditions. It reveals the presence of small size visible pores as
compared to the atmospheric condition. Further, Fig. 3 shows the source images of Cu composites and the respective segmented
images before / after sintering. The raw and segmented images of Cu-5W before and after sintering are shown in Fig. 4 and
source and segmented images of Cu-20W before and after sintering are shown in Fig.5. Table 1 reports the measurements of
pores of various composites before and after sintering. It seems that the pore area and perimeter has been reduced after sintering
process of all Coppertungsten (CuW) types. Similarly the source and segmented images generated during the extrusion
process of the composite preforms of Cu-0W, Cu-5W,Cu-10W and Cu20W are exhibited in Fig.6,7 and 8 respectively. Table 2
summarized the pore properties of SEM images of before and after extrusion process. The intensity of seed and threshold used
during region growing are reported in Table-3.

Fig. 1: Sintered preforms

Fig. 2: (a) SEM image generated during sintering at

Normal atmosphere (850C)

Fig. 2: (b) SEM image generated during sintering at

argon atmosphere (800C)

All rights reserved by www.grdjournals.com

150

Region Growth Based Segmentation to Improve the Porosity of Cu - (520%) W Composite Preforms
(GRDJE/ Volume 1 / Issue 12 / 025)

Fig. 3: Raw and segmented images of Cu W composite before and after sintering

Fig. 4: Raw & segmented images of Cu 5W composite before and after sintering

All rights reserved by www.grdjournals.com

151

Region Growth Based Segmentation to Improve the Porosity of Cu - (520%) W Composite Preforms
(GRDJE/ Volume 1 / Issue 12 / 025)

Fig. 5: Raw & segmented images of Cu-20W composite before sintering and after sintering

Fig. 6: Raw & segmented images of Cu-W composite before and after Extrusion

All rights reserved by www.grdjournals.com

152

Region Growth Based Segmentation to Improve the Porosity of Cu - (520%) W Composite Preforms
(GRDJE/ Volume 1 / Issue 12 / 025)

Fig. 7: Raw & Segmented images of Cu-5W composite before and after Extrusion

Fig. 8: Raw & Segmented images of Cu-10W composite before and after extrusion

Region Property
No. of regions
Area

Table 1: Cu-W nano composites at room and argon atmosphere

Sintering at normal atmosphere (850 C) Sintering at argon atmosphere (800 C)
42
6
38764
33828

All rights reserved by www.grdjournals.com

153

Region Growth Based Segmentation to Improve the Porosity of Cu - (520%) W Composite Preforms
(GRDJE/ Volume 1 / Issue 12 / 025)

Perimeter
Texture properties
Average Intensity
Average contrast
R
Third Moment
Uniformity
Entropy Source Image
Entropy Segmented Image
No. of regions
Area
Perimeter
Texture properties
Average Intensity
Average contrast
R
Third Moment
Uniformity
Entropy Source Image
Entropy Segmented Image
No. of regions
Area
Perimeter
Texture properties
Average Intensity
Average contrast
R
Third Moment
Uniformity
Entropy Source Image
Entropy Segmented Image

10007

5676.6

39.5393
92.2993
0.1158
23.0481
0.7380

43.2439
95.6931
0.1234
23.7308
0.7183

7.0296
0.6223
Nano Composite Cu-5W
20
49029
8911.8

5.6588
0.6568

50.0096
101.2496
0.1362
24.4334
0.6847
6.2649
0.7141
Nano Composite Cu-20W
26
30931
7449.3

59.1026
107.6013
0.1511
24.3570
0.6439
7.1907
0.7811

29
9271
4513.8

22
17818
11337

31.5496
42.1617
83.9629
94.7292
0.0978
0.1213
20.8052
23.5538
0.7832
0.7240
7.2848
7.2744
0.5400
0.6469
Table -2 Composites Before or after extrusion
Nano Composite Cu-W
Region Property
Before Extrusion After Extrusion
No. of regions
112
31
Area
172781
52882
Perimeter
26003.13
9784.254
Texture properties
Average Intensity
89.3071
27.12996
Average contrast
121.645
78.62636
0.18538
0.086819
R
17.3829
19.08491
Third Moment
0.54487
0.809855
Uniformity
Entropy Source Image
5.85128
6.865375
Entropy Segmented Image
0.934268
0.488936
Nano Composite Cu-5W
No. f regions
8
64
Area
187629
92920
Perimeter
69767
11231
Texture properties
97.66092
48.22764
Average Intensity
123.9592
99.86062
Average contrast
0.19114
0.132967
R
14.10238
24.3142
Third Moment
0.527385
0.693283
Uniformity
Entropy Source Image
5.988404
6.883518
Entropy Segmented Image
0.960122
0.699644
Nano Composite Cu-10W
No. of regions
37
81
Area
139370
84253
Perimeter
16959
16456

All rights reserved by www.grdjournals.com

154

Region Growth Based Segmentation to Improve the Porosity of Cu - (520%) W Composite Preforms
(GRDJE/ Volume 1 / Issue 12 / 025)

Texture properties
72.52929
43.84595
Average Intensity
115.0412
96.2198
Average contrast
0.16911
0.124634
R
22.37626
23.8213
Third Moment
0.592942
0.71524
Uniformity
Entropy Source Image
6.85311
7.357052
Entropy Segmented Image
0.861415
0.662135
Nano Composite Cu-20W
No. of regions
15
22
Area
151275
49024
Perimeter
8207.9
6247.5
Texture properties
78.72474
25.51249
Average Intensity
117.8016
76.51665
Average contrast
0.175879
0.082602
R
20.81861
18.36574
Third Moment
0.573173
0.819922
Uniformity
Entropy Source Image
6.886508
7.540112
Entropy Segmented Image
0.891696
0.469
Table 3: Intensity of seed and threshold during region growing
S.No
Type
Before sintering After sintering
Cu-0W
97, 27
73, 20
Cu-5W
72, 21
84, 32
Cu-20W
100, 30
70, 25
Cu-0W
30, 15
40, 30
Cu-5W
35, 20
35, 20
Cu-10W
20, 20
20, 20
Cu-20W
20, 20
55, 20

V. CONCLUSION
The pore area detection and rectification is a major task in the manufacturing section of mechanical products. This region based
segmentation mentioned in this paper helps to identify the pore areas of Cu-W composites without destructing the materials. The
SEM image based Analysis is helpful to detect the morphological characteristics of material defects. . Further the soft computing
approach like Genetic algorithm may be applied to identify the optimal features of SEM images while characterizing the
composites in the manufacturing section of mechanical parts in future.

REFERENCES
Rafael C.Conzalez, Richard E.Woods, Steven L.Eddins, Digital Image processing, Pearson Education, 2008.
N. Leema, P.Radha, S.C. Vettivel, H. Khanna Nehemiah, Characterization, pore size measurement and wear model of a sintered CuW nano composite
using radial basis functional neural network, Materials and Design 68 (2015) pp.195206.
[3] Shwetabh Singh, Microscopic Image Analysis of Nanoparticles by Edge Detection using Ant Colony Optimization, IOSR Journal of Computer
Engineering 11(3) (2013) pp. 84-89.
[4] Azmi Tawfik Alrawi , Saad Jasim Mohammed, Determination The Porosity of CdS Thin Film by Seed Filling Algorithm, International Journal on Soft
Computing 3(3) (2012) pp:1-12
[5] Meena Sundari P.,Meenambal T.,and P. D. Arumairaj, Artificial Neural Networks based prediction of removal of basic dye from its aqueous solution
using activated carbon, Proceedings of the 3rd World Conference on Applied Sciences, Engineering and Technology 27-29, Kathmandu, Nepal, ISBN 13:
978-81-930222-0-7, (2014) pp 313-317.
[6] Gary Chinga, Structural studies of LWC paper coating layers using SEM and image analysis Techniques, Ph.D. thesis, Norwegian University of Science
and Technology, February 2002.
[7] Manuel F.M.Costa, Application of image processing to the characterization of nano structures, Advanced Material Science 6 (2004) pp.12-20.
[8] M. G. Cortina-Januchs, J. Quintanilla-Dominguez,, A. Vega-Corona, A. M. Tarquis,D. Andina, Detection of pore space in CT soil images using artificial
neural networks, Biogeo sciences, 8, (2011) pp.279288.
[9] Aref Naimzad, Yousef Hojjat, Mojtaba Ghodsi, Comparative Study on Mechanical and Magnetic Properties of Porous and Nonporous Film-shaped
Magneto rheological Nanocomposites Based on Silicone Rubber International Journal of Innovative Science and Modern Engineering, 2(8) (2014)
pp:11-20.
[10] Ahmet H. Aydilek, A.M.ASCE; Seyfullah H. Oguz,, and Tuncer B. Edil, M.ASCE , Digital Image Analysis to Determine Pore Opening Size Distribution
of Nonwoven Geotextiles:, Journal of computing in civil engineering, Vol:16,issue:4, (2002) pp: 280-290.
[11] Shawn Zhang, Robert E. Klimentidis, Patrick Barthelemy, Study of membrane morphology of SEM image of polymer nanocomposite membrane by
digital Image processing, International Symposium of the Society of Core Analysts held in Austin, Texas, USA, 2011.
[1]
[2]

All rights reserved by www.grdjournals.com

155